Reduce Image Size Below 100KB

A typical phone photo is 3 to 6 MB, which is 3,000 to 6,000 KB. Getting to 100 KB means removing about 97 per cent of the data. That sounds severe, but it is achievable because most of that data is detail you cannot see on a small screen.

At 100 KB, a portrait of 1,000 by 1,300 pixels at a medium quality looks fine for a form. A full-width landscape photo at 1,600 pixels wide is also possible, though with softer fine detail such as hair or leaves.

A scanned page with text can reach 100 KB at about 1,200 pixels wide, though small print will lose crispness. For documents, check the text is readable before you submit.

Step one is to shrink the pixel dimensions. Most portals do not need more than 1,200 pixels on the long edge. This alone often takes a 4 MB photo to about 400 KB.

Step two is to lower the quality to about 70 to 80. The file drops again, often to between 80 and 150 KB, depending on the picture.

Step three is to let the tool fine-tune. It adjusts the quality in small steps until the file is just under 100 KB, so you do not lose more quality than necessary.

Photos with soft backgrounds compress well. A portrait against a plain wall may drop below 100 KB with only a small resize.

Photos with fine detail, such as foliage or crowds, are harder. They need smaller dimensions to reach the same size.

Screenshots with flat colours and sharp text compress well as PNG, but a JPEG at moderate quality is often smaller. Check the preview for ringing around text.

Scans with grey backgrounds can hide noise, which inflates the file. Scanning at 150 DPI instead of 300 usually helps a lot.

Open the downloaded file and zoom to 100 per cent. Check the face, text or numbers you care about. If they look blurry, raise the target a little and run again.

Check the format and name. JPEG is accepted almost everywhere. Use a short name without spaces.

If the portal still rejects the file, read the error. It may be about dimensions, aspect ratio or format rather than size.

A certificate scanned at 300 DPI on A4 paper is about 2,480 by 3,508 pixels and can be 2 to 3 MB as a JPEG. The portal wants it under 100 KB.

The first step is to reduce the long edge to about 1,600 pixels, which keeps the text readable on screen. That alone gives roughly 450 KB. The second step is to lower the quality to about 65, which brings it to about 95 KB. The preview shows the text still sharp at 100 per cent zoom.

If small print becomes hard to read, raise the target to 120 KB if the portal allows, or crop away blank margins before compression so that the pixels are used on the content.

JPEG is the safest choice for photos and scans. It is accepted almost everywhere and compresses well at moderate quality.

PNG is lossless and best for flat graphics and screenshots with text. It is usually larger than JPEG for photos, so it is rarely the right choice when you have a tight KB limit.

WebP gives smaller files than JPEG at the same visual quality, but not every portal accepts it. If the form does not list WebP as allowed, use JPEG.

This tool saves JPEG when you set a size target, which matches what most portals expect.

For faces, keep the face large in the frame. A tightly cropped portrait uses its pixels on the face, so you can reduce the overall dimensions less.

For text, avoid going below about 1,200 pixels on the long edge for a full page. Below that, small characters begin to merge.

For both, do not apply sharpening before compression. Sharpening adds fine detail that the encoder then has to store, which inflates the file for no visible gain at form sizes.

Applications often require several documents: a photo, a signature, an ID scan and a certificate. Each has its own limit. Make a short list with the limit for each file before you start.

Process the files in order from the strictest limit to the loosest. That way you finish the hardest jobs while you are fresh, and the easy ones go quickly.

Name each result clearly, for example photo-100kb.jpg and id-scan-100kb.jpg. If you upload to the wrong field, the portal may reject the file or, worse, accept it and cause confusion later.

Keep all the originals in a separate folder. If a portal changes the limit, you can go back to the full-quality file.

If the text on a scan becomes hard to read at 100 KB, try converting the page to black and white before compression. A two-tone scan has far less colour data and can be much smaller at the same pixel size.

Crop the margins. Blank borders use pixels and bytes for nothing. A tight crop lets you keep more resolution on the text.

Scan at a lower DPI to start with. 150 DPI is plenty for on-screen reading, and it halves each dimension compared with 300 DPI.

If none of this is enough, check whether the portal accepts PDF. A PDF with a single compressed image may be handled differently, and the limit may be higher.

Is the file under the limit? Check the number in the preview, not only in your file manager.

Is the format correct? JPEG is safest unless the portal says otherwise.

Are the dimensions inside the allowed range? Look at the width and height.

Is the content readable at 100 per cent zoom? Faces, numbers and text should be clear.

Is the file name simple? Use letters, numbers and a hyphen, with no spaces or special characters.

Renaming a file from .png to .jpg does not change its format or size. The portal may still see a PNG inside and reject it. Use the tool to convert properly.

Taking a screenshot of a photo and uploading that screenshot lowers quality and may add unwanted interface elements. Use the original file.

Compressing the same file many times in a row degrades it with each pass. Always go back to the original for each new target.

Ignoring the pixel dimensions. A form may accept the file size but reject a picture that is too small or too large in pixels. Check both numbers.

It is large enough to hold a clear portrait or a readable page, and small enough to keep storage and transfer costs low for the site owner.

It is also a round number that is easy to state in an instruction, which is why so many forms use it.

For users, it is the point where most photos need a conscious step to shrink. Below 100 KB the effort rises, and above it the effort is usually minimal.

Start with the dimensions that the portal actually shows. If a form displays the photo at 200 by 250 pixels, anything above about 600 by 750 is wasted. Resize to a comfortable margin above what is displayed, then compress.

Remove metadata. Phone photos include location, camera and edit data that can add tens of kilobytes. Re-encoding through the tool strips most of it, which also protects your privacy when you share the file.

Avoid heavy filters before compression. Filters that add grain, glow or texture create detail the encoder has to keep, which makes the file larger. Apply light edits only.

Compare before and after at the size the picture will be viewed. If you cannot see a difference at that size, the extra kilobytes were not doing anything useful.

If you must keep more detail, pick the area that matters and crop to it. A tight crop on a face or on a block of text is better than a wide shot compressed to the same limit.

Some portals count size in kilobytes of 1,000 bytes rather than 1,024. A file shown as 100 KB on one system may be 102 KB on another. Aim a few kilobytes under the limit to be safe.

Some check the colour profile. A file saved with an unusual profile may be flagged. This tool outputs a standard profile that portals accept.

Some check the file extension against the contents. A file named .jpg that is really a different format may be refused. Re-encode through the tool rather than renaming.

Some have a time-out on slow connections. A smaller file uploads faster, so even a file inside the limit may be worth shrinking further on a weak network.

Government portals and visa application systems almost always cap ID photos at 100 KB. A typical smartphone camera produces a JPEG between 2 MB and 5 MB, so you need to shed roughly 95 % of the file size while keeping your face clearly identifiable. The key is to compress in two stages: first reduce pixel dimensions to the portal's specified size (commonly 413 × 531 px for a 35 × 45 mm print at 300 DPI, or 600 × 800 px for digital submissions), then lower JPEG quality until the file sits comfortably under 90 KB — leaving a 10 KB buffer for the reasons described in the rejection section above.

Scanned documents follow the same two-step logic but with different starting sizes. A flatbed scan at 300 DPI of an A4 certificate produces a PNG around 8–15 MB. The fastest route is to save it as JPEG at 85 % quality after scaling the long edge to 2,000 px — that alone usually lands below 500 KB. A second pass at 70 % quality with the long edge at 1,600 px typically drops you under 100 KB without making text unreadable. Run a quick zoom to 100 % in your browser after compressing: if individual letters are still distinct, the document will pass an automated legibility check.

The three most common failure points for ID photos are: (1) a white border required by the portal was accidentally cropped, forcing you to re-scan and restart the compression chain; (2) the EXIF orientation tag was stripped, causing the portal to display the photo sideways even though it looks correct on your device; (3) the background is slightly off-white (RGB values like 248, 248, 248 instead of 255, 255, 255), which optical character recognition systems flag as a dirty background. Fixing any of these requires editing the image before compressing, not after — re-compressing an already-compressed JPEG accelerates quality loss without fixing the underlying issue.

For printed certificates that include fine-print text below 8 pt, prefer PNG over JPEG even if it costs a few extra kilobytes. JPEG's block artefacts appear first around sharp high-contrast edges — exactly where small letters live. If PNG pushes you over 100 KB, convert to WebP at 80 % quality instead. WebP encodes the same visual information roughly 30 % smaller than JPEG at equivalent settings, and all modern government portal browsers (Chrome 80+, Firefox 65+, Safari 14+) accept it.

The most frustrating outcome is reaching the file size limit but losing readability in the process. This happens when you apply a single large compression step rather than several small ones. Each lossy JPEG encode multiplies the artefacts from the previous encode — a pattern called generation loss. If your starting file is already a JPEG (common for phone photos), decode it once, make all your edits, then encode to the final JPEG exactly once. Tools that silently re-encode on every save, such as some online editors, can push a 4 MB photo through five encode cycles before you download it, adding 15–25 % more artefact noise for the same output file size.

When the file stubbornly stays above 100 KB after quality reduction, the limiting factor is usually image dimensions rather than quality. A 4,032 × 3,024 px photo at JPEG quality 20 is still around 400 KB because the raw pixel count is so large. Halving the long edge to 2,016 px at quality 60 almost always produces a better-looking file at a smaller size — because quality artefacts at 60 are far less visible than at 20, and the dimension reduction does most of the size work. A useful mental model: every time you halve linear dimensions you cut file size by roughly 75 %; every 10-point drop in JPEG quality (from 80 to 70, for example) cuts file size by roughly 20–30 %.

Screenshots of documents — captured from a PDF viewer or e-mail client — introduce a hidden problem: the operating system renders the screen at 2× or 3× pixel density on retina displays, so a screenshot that looks like it shows an A4 page is actually 5,000–6,000 px tall. Before compressing, check the actual pixel dimensions in the file details. If the height exceeds 3,000 px, resize to 50 % first. Skipping this step wastes cycles because the compressor has to process four times as many pixels to reach the same output size.

File format confusion accounts for a surprising share of failures. A PNG screenshot of black text on a white background is already highly compressible — converting it to JPEG at 80 % quality routinely cuts size by 85 % with no visible degradation. Conversely, a colour photo saved as PNG will be two to four times larger than the same photo as JPEG, because PNG uses lossless compression that is not designed for continuous-tone photographic data. The rule: text and diagrams with sharp edges → PNG or WebP lossless; photographs and portraits → JPEG or WebP lossy.